CA2114052A1 - Method of detecting scene cuts - Google Patents
Method of detecting scene cutsInfo
- Publication number
- CA2114052A1 CA2114052A1 CA 2114052 CA2114052A CA2114052A1 CA 2114052 A1 CA2114052 A1 CA 2114052A1 CA 2114052 CA2114052 CA 2114052 CA 2114052 A CA2114052 A CA 2114052A CA 2114052 A1 CA2114052 A1 CA 2114052A1
- Authority
- CA
- Canada
- Prior art keywords
- frame
- gradual
- cuts
- scene
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000000694 effects Effects 0.000 description 3
- 101100008046 Caenorhabditis elegans cut-2 gene Proteins 0.000 description 1
- 101100008049 Caenorhabditis elegans cut-5 gene Proteins 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
- H04N5/147—Scene change detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B27/00—Editing; Indexing; Addressing; Timing or synchronising; Monitoring; Measuring tape travel
- G11B27/02—Editing, e.g. varying the order of information signals recorded on, or reproduced from, record carriers
- G11B27/031—Electronic editing of digitised analogue information signals, e.g. audio or video signals
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11B—INFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
- G11B27/00—Editing; Indexing; Addressing; Timing or synchronising; Monitoring; Measuring tape travel
- G11B27/10—Indexing; Addressing; Timing or synchronising; Measuring tape travel
- G11B27/19—Indexing; Addressing; Timing or synchronising; Measuring tape travel by using information detectable on the record carrier
- G11B27/28—Indexing; Addressing; Timing or synchronising; Measuring tape travel by using information detectable on the record carrier by using information signals recorded by the same method as the main recording
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/142—Detection of scene cut or scene change
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/179—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a scene or a shot
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
- H04N19/87—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression involving scene cut or scene change detection in combination with video compression
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
- H04N5/144—Movement detection
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Television Signal Processing For Recording (AREA)
- Image Analysis (AREA)
- Studio Circuits (AREA)
Abstract
METHOD FOR DETECTING SCENE CUTS
ABSTRACT
A method for operating a computer or similar apparatus so as to automatically detect scene cuts in a motion picture. The motion picture comprises a plurality of frames in a sequential order. The preferred embodiment of the present invention utilizes the correlation between successive frames to detect abrupt cuts. Gradual cuts are preferably detected by examining a function consisting of the standard deviation of the pixels of each frame as a function of the frame number.
ABSTRACT
A method for operating a computer or similar apparatus so as to automatically detect scene cuts in a motion picture. The motion picture comprises a plurality of frames in a sequential order. The preferred embodiment of the present invention utilizes the correlation between successive frames to detect abrupt cuts. Gradual cuts are preferably detected by examining a function consisting of the standard deviation of the pixels of each frame as a function of the frame number.
Description
:` 2~ 2 M~THOD F~ 2ETEcTlNG SCENE CU~
= ~1~
., The present invention relates to video processing systems, and more particul~ly, to an improved method for automatically detect~ng the s~art of a n~w scene in a video seguence.
;~
Backgr,~ Lof the InYention ~ ~:
A typical motion pictwre consists of a sequence of scenes. Within a scene, each frame differs only slightly from the preceding frame. The change from one scene to the next will be referred to as "cut" in the following discussion. Scene cuts may be abrupt or gradual. Gradual scene CUt5 are associated with special effects such a5 camera pans, dissolves, fades to black or fade from black.
~:
One method for swmmarizing a motion picture is to provide a log of the scenes. A scene log is a list of the scenes and the location of the start of the scene from some reference point such as the start of the motion picture. A typical frame from the scene may be used to summarize the scene.
.
Among other things, such logs are useful in video editing. The log provides a means for an editor to quickly find a particular scene.
~ .
Prior art schemes for automating the logging of motion pictures rely on the difference between sequential frames to determine where a cut has occurred. These methods compute the average value of some parameter such as the absolute value of the difference of corresponding pixels in two sequential frames. Such methods do not detect gradual cuts with sufficient reliability to be useful in automating the cut logging process. In addition, there are some situations in which even abrupt cuts are ` not detected by these methods.
~`:
Broadly, it is the object of the present invention to provide an improved method for detecting scene cuts.
., 2 ~ 2 It is a further object of the present invention to provide an automated method for detecting both gradual scene and abrupt cuts. ~
These and other objects of the present invention will become apparent to those ~ ; -skilled in the ar~fiom the following detailed description~-the-invention and~theaccompanying drawings.
Summary~oftheInvention ; ~
'", ~' The present invention comprises a method for operating a computer or similar apparatus so as to automatically detect scene cuts in a motion picture. The motion ;~
picture comprises a plurality of frames in a sequential order. The preferred embodiment of the present invention utilizes the correlation between successive~ ~
frames to detect abrupt cuts. Gradual cuts are preferably detected by examining a ~ ;
function consisting of the standard deviation of the pixels of each frame as a function of the frame number. Gradual cuts are characterized by excursions in this function.
A gradual scene cut may either manifest itself by a peak in this function or by a ramp connecting two regions of relatively constant values.
~0 ' ~ .'' ,~
. ..... . . . ..
Figure I is a flow chart for a program according to the present invention for 25 finding the cuts in a motion picture.
12etailed Description of the Invention . , The present invention utilizes the correlation of the pixels in successive 30 frames to detect abrupt scene cuts. The method by which more gradual scene cuts are detected will be explained in more detail below. Denote the pixels of the nth frame of a motion picture by nI(x,y). That is, nl(x,y) is the intensity of the pixel at some point (x,y) in the frame. For the purposes of this discussion, it will be assumed that nI(x,y) I ~ x:~
is defined on a regular grid having a total of N grid points. For two frames, (n-k) and n, the method of the present invention computes a correlation factor, Pn k~ where E[(n-k)I(x y)*"I(x,y)]--E[(n-k)I(x,y)]*E[ I(x,y)] (1) ~SD ~(n-5 where E[(nI(x,y)3 = N ~,~, nI(x,y) (2 and ~Sn =~E[nl2(X~Y)]--E2[nI(x,y)] (3) ~ ':
In the preferred embodiment of the present invention, k=l . Wi~in any given scene, Pn I is expected to be near one. If an abrupt scene cut occurs at frame n, then Pn decrease to near zero. The present invention tracks the value of Pn I fr~m frame to frame and defines an abrupt scene cut as having occurred at frame n if Pn I is below 15 some predetermined value.
In practice, the threshold value may be found by examining a few scene changes. However, it will be apparent to those skilled in the art that fixed thresholds may be used. The cut-offcan also be defined in terms of the previously observed 20 values Of Pn 1. In such a system, an abrupt scene cut would be defined to have occurred at frame n if Pn I is less than the average value of Pn I over the previous M
frames where M is some integer greater than 0.
While the above method finds abrupt scene cuts with a high degree of 25 reliability, it is not a reliable method for fnding gradual scene cuts, since the amount of change between successive frames in a gradual cut is too small to substantially change the value of Pn 1. In principle, the method discussed above can be extended to find gradual scene cuts by increasing the "distance" between the frames for which Pn k is computed, i.e., choosing k>l. In such a system, abrupt cuts would be detected30 using Pn 1~ and gradual cuts would be detected using Pn k with k set to some predetermined value. While such a scheme is capable of detecting more of the ,, . ' 211 !10S2 : ~
gradual cuts, it requires some knowledge of the correct value for k. In addition, -further analysis is needed to determine the precise location of the cut.
In the preferred embodiment of the present invention, an alternative method is used to detect gradual cuts. As discussed above, the computation of Pn k requires the : :~
computation of c~n. The preferred embodiment of the present invention uses the behavior of ~n to detect gradual scene cuts. The value of ~n for any frame measures the variation of the pixel values about the mean pixel value in the frame. Within any ;
given scene, the values of ~n from frame to frame remain roughly constant. When a ~;~
10 scene slowly changes, the value ~n changes from the value representative of the ~ -previous scene to the value representative of the new scene. This change takes place over a number of frames.
. .
The specific manner in which C~n changes will depend on the starting and 15 ending scene, and the nature of the special effect. For example, in the case of a fade to black followed by a fade from black to the new scene, ~n would be expected todecrease during the fade to black and then increase to the value representative of the new frame. In practice, the ~n curve has considerably more noise than the Pn k curve;
hence, the curve is smoothed in the preferred embodiment of the present invention to 20 allow automated determination of the gradual cuts. In the preferred embodiment of the present invention, the previous 4 points are averaged to generate the smoothed version of CJn ~ ..
The excursions of the ~n curve during a gradual scene cut are of two possible 25 types. In the first type, the ~n curve goes through a peak. This is similar to the fade to black effect discussed above. In the second type, the an curve exhibits a "ramp"
connecting the two regions of fairly constant ~n values.
A gradual cut is defined as having begun when the absolute value of the slope 30 of the smoothed ~n curve increases beyond a predetermined threshold value. The increased slope value must be present for a predetermined first time interval, i.e., number of frames, before the gradual cut is regarded as valid. When the absolutevalue of the slope has remained below a predetermined threshold for a second time interval, the gradual cut is deemed to have been completed. Between the beginning ~ ~, .` ` 1~ . ~ ' ~; ' ; , ~ ., j i 9 ~ 2 and end of the gradual cut, the slope of the ~sn curve will, in general, go to zero briefly as the ~n curve passes through an inflection point if a peak is present; hence, the gradual cut is not defined to have been completed until the absolute value of the slope of the (Sn curve remains below the threshold for some predetermined time interval.
S :: ~
Re~er ~w to-~i~e l which-is a-~wc-~ of-~efle eut det~cting-program according to the present invention. In the preferred embodiment ofthe pTesent invention, the value of k in Eq. (1) is set to one. To sirnplify the notation, Pn I has been replaced by p in the following discussion. After initializing various variables as 10 shown at 10, the program acquires the next frame of the motion picture and computes the values for p and the average value of ~J as shown at 12. The program then tests the value of p to deterrnine if an abrupt cut has occurred as shown at 14. An abrupt cut is deemed to have occurred if the value of p is less than a predetermined threshold. It is possible for an abrupt cut to take more than one frame. Hence, a flag l S is set as shown at 16 to indicate that p has crossed the threshold. The flag will be reset when p once again retums to a value above the threshold. This flag is used by blocks 18 and 20 to determine if the current value of p being less than the threshold is a new cut or just a continuation of the previously detected abrupt cut. If the cut is completed, the flag is reset as shown at 22.
If no abrupt cut is in progress, the program goes on to examine the data to determine if a gradual cut is in progress. As noted above, the program uses the average value of <~ to detect gradual cuts. The curve consisting of the average value of ~ as a function of the frame number will be referred to as the ~s curve. The 25 program also keeps track of the slope of the cJ curve. Changes in the slope of the a curve are used to detect possible events in traversing a gradual cut. As noted above, a gradual cut may be evidenced either by a peak in the ~ curve or by a ramp connecting two ~ values representative of ~s in the frarnes on each side of the gradual cut. In the later case, there will be no change in the sign of the slope of the ~ curve during the 30 gradual cut.
Referring again to Figure 1, if no abrupt cut is in progress, the program tests for a gradual cut being in progress as shown at 24. If a gradual cut is not in progress, the program checks for the beginning of a gradual cut as shown at 26. A gradual cut 2 ~ 2 is deemed to have begun if ~e slope of ~e ~ curve has changed and ~e change in 6is greater than a predetennined threshold value. If these conditions are met, a set of flags is initialized as shown at 28. The flag values indicate that a gradual cut has begun and no peak has yet been detected in ~e 6 curve.
If ~a~al cut-was already-in progres~-a~ ate~by-the flags set at 28, the program determines if the current f~ame is the end of the gradual cut as shown at 30. -If the sign of the slope of the c~ curve has changed, the program checks to see if the program has already passed through an inflection point as indicated by one of the 10 above-described flags having a value different from its initialization value. If the initial value is still present, the program assumes that the change in sign is due to the presence of an inflection point in the ~ curve. The flag is set accordingly as shown at 32. If the ~ curve has already passed through an inflection point, the program assumes that the gradual cut has been completed.
If the sign of the slope of the ~ curve has not changed but the change in the c~curve from the last frame is less than some predetennined threshold, the programassumes that the gradual cut has been completed, provided the excursions in the ~
curve from one scene to the next were sufficiently large. This case is encountered 20 when there is no peak in the a curve between the scenes.
The a curve, even with smoothing, has a significant amount of noise. To distinguish a "ramp" between two values in the ~s curve from noise, the program keeps track of certain reference values on the <s curve. The value of the a curve at the 25 potential beginning of a ramp is stored. At the end of the ramp, the difference between the ending value and the reference values is used to eliminate "noise events".
While the preferred embodiment of the present invention uses the standard deviation of the pixel values in each frame for detecting gradual cuts, it will be 30 apparent to those skilled in the art that other statistical measures or functions thereof may be used. In this regard, it should be noted that ~ is a function of the variance of the pixel values which in turn is the average value of the second moment of the pixel values. Hence, any function based on one of the moments of the pixel distribution may also be utilized.
~ ~ ~ r :~
2 ~ 2 -7~
While the above-described embodiments of the present invention compute the correlation and standard deviation using all of the data in each frame, it w~ll be apparent to those skilled in the art that a subset of the pixel values may be used to determine these parameters. Any representative subset may be used. Embodiments ~ -based on such subsets generate a lower computational load; however, such embodiments are expected to yield ~ values having more "noise".
The present invention is preferably practiced on a general purpose computer.
However, it will be apparent to those skilled in the art that special purpose computing hardware may also h utilized to improve the speed with which the method operates.
For example, hardware that computes the correlation of t~vo frames and the variance of the pixels in a given frame can substantially improve ~e speed with which the ~ `
method of the present invention operates. -Various modifications to the present invention will become apparent to those skilled in the art from the foregoing description and accompanying drawings.
Accordingly, the present invention is to be limited solely by the scope of the following claims. ~ ; -~ ~ S
= ~1~
., The present invention relates to video processing systems, and more particul~ly, to an improved method for automatically detect~ng the s~art of a n~w scene in a video seguence.
;~
Backgr,~ Lof the InYention ~ ~:
A typical motion pictwre consists of a sequence of scenes. Within a scene, each frame differs only slightly from the preceding frame. The change from one scene to the next will be referred to as "cut" in the following discussion. Scene cuts may be abrupt or gradual. Gradual scene CUt5 are associated with special effects such a5 camera pans, dissolves, fades to black or fade from black.
~:
One method for swmmarizing a motion picture is to provide a log of the scenes. A scene log is a list of the scenes and the location of the start of the scene from some reference point such as the start of the motion picture. A typical frame from the scene may be used to summarize the scene.
.
Among other things, such logs are useful in video editing. The log provides a means for an editor to quickly find a particular scene.
~ .
Prior art schemes for automating the logging of motion pictures rely on the difference between sequential frames to determine where a cut has occurred. These methods compute the average value of some parameter such as the absolute value of the difference of corresponding pixels in two sequential frames. Such methods do not detect gradual cuts with sufficient reliability to be useful in automating the cut logging process. In addition, there are some situations in which even abrupt cuts are ` not detected by these methods.
~`:
Broadly, it is the object of the present invention to provide an improved method for detecting scene cuts.
., 2 ~ 2 It is a further object of the present invention to provide an automated method for detecting both gradual scene and abrupt cuts. ~
These and other objects of the present invention will become apparent to those ~ ; -skilled in the ar~fiom the following detailed description~-the-invention and~theaccompanying drawings.
Summary~oftheInvention ; ~
'", ~' The present invention comprises a method for operating a computer or similar apparatus so as to automatically detect scene cuts in a motion picture. The motion ;~
picture comprises a plurality of frames in a sequential order. The preferred embodiment of the present invention utilizes the correlation between successive~ ~
frames to detect abrupt cuts. Gradual cuts are preferably detected by examining a ~ ;
function consisting of the standard deviation of the pixels of each frame as a function of the frame number. Gradual cuts are characterized by excursions in this function.
A gradual scene cut may either manifest itself by a peak in this function or by a ramp connecting two regions of relatively constant values.
~0 ' ~ .'' ,~
. ..... . . . ..
Figure I is a flow chart for a program according to the present invention for 25 finding the cuts in a motion picture.
12etailed Description of the Invention . , The present invention utilizes the correlation of the pixels in successive 30 frames to detect abrupt scene cuts. The method by which more gradual scene cuts are detected will be explained in more detail below. Denote the pixels of the nth frame of a motion picture by nI(x,y). That is, nl(x,y) is the intensity of the pixel at some point (x,y) in the frame. For the purposes of this discussion, it will be assumed that nI(x,y) I ~ x:~
is defined on a regular grid having a total of N grid points. For two frames, (n-k) and n, the method of the present invention computes a correlation factor, Pn k~ where E[(n-k)I(x y)*"I(x,y)]--E[(n-k)I(x,y)]*E[ I(x,y)] (1) ~SD ~(n-5 where E[(nI(x,y)3 = N ~,~, nI(x,y) (2 and ~Sn =~E[nl2(X~Y)]--E2[nI(x,y)] (3) ~ ':
In the preferred embodiment of the present invention, k=l . Wi~in any given scene, Pn I is expected to be near one. If an abrupt scene cut occurs at frame n, then Pn decrease to near zero. The present invention tracks the value of Pn I fr~m frame to frame and defines an abrupt scene cut as having occurred at frame n if Pn I is below 15 some predetermined value.
In practice, the threshold value may be found by examining a few scene changes. However, it will be apparent to those skilled in the art that fixed thresholds may be used. The cut-offcan also be defined in terms of the previously observed 20 values Of Pn 1. In such a system, an abrupt scene cut would be defined to have occurred at frame n if Pn I is less than the average value of Pn I over the previous M
frames where M is some integer greater than 0.
While the above method finds abrupt scene cuts with a high degree of 25 reliability, it is not a reliable method for fnding gradual scene cuts, since the amount of change between successive frames in a gradual cut is too small to substantially change the value of Pn 1. In principle, the method discussed above can be extended to find gradual scene cuts by increasing the "distance" between the frames for which Pn k is computed, i.e., choosing k>l. In such a system, abrupt cuts would be detected30 using Pn 1~ and gradual cuts would be detected using Pn k with k set to some predetermined value. While such a scheme is capable of detecting more of the ,, . ' 211 !10S2 : ~
gradual cuts, it requires some knowledge of the correct value for k. In addition, -further analysis is needed to determine the precise location of the cut.
In the preferred embodiment of the present invention, an alternative method is used to detect gradual cuts. As discussed above, the computation of Pn k requires the : :~
computation of c~n. The preferred embodiment of the present invention uses the behavior of ~n to detect gradual scene cuts. The value of ~n for any frame measures the variation of the pixel values about the mean pixel value in the frame. Within any ;
given scene, the values of ~n from frame to frame remain roughly constant. When a ~;~
10 scene slowly changes, the value ~n changes from the value representative of the ~ -previous scene to the value representative of the new scene. This change takes place over a number of frames.
. .
The specific manner in which C~n changes will depend on the starting and 15 ending scene, and the nature of the special effect. For example, in the case of a fade to black followed by a fade from black to the new scene, ~n would be expected todecrease during the fade to black and then increase to the value representative of the new frame. In practice, the ~n curve has considerably more noise than the Pn k curve;
hence, the curve is smoothed in the preferred embodiment of the present invention to 20 allow automated determination of the gradual cuts. In the preferred embodiment of the present invention, the previous 4 points are averaged to generate the smoothed version of CJn ~ ..
The excursions of the ~n curve during a gradual scene cut are of two possible 25 types. In the first type, the ~n curve goes through a peak. This is similar to the fade to black effect discussed above. In the second type, the an curve exhibits a "ramp"
connecting the two regions of fairly constant ~n values.
A gradual cut is defined as having begun when the absolute value of the slope 30 of the smoothed ~n curve increases beyond a predetermined threshold value. The increased slope value must be present for a predetermined first time interval, i.e., number of frames, before the gradual cut is regarded as valid. When the absolutevalue of the slope has remained below a predetermined threshold for a second time interval, the gradual cut is deemed to have been completed. Between the beginning ~ ~, .` ` 1~ . ~ ' ~; ' ; , ~ ., j i 9 ~ 2 and end of the gradual cut, the slope of the ~sn curve will, in general, go to zero briefly as the ~n curve passes through an inflection point if a peak is present; hence, the gradual cut is not defined to have been completed until the absolute value of the slope of the (Sn curve remains below the threshold for some predetermined time interval.
S :: ~
Re~er ~w to-~i~e l which-is a-~wc-~ of-~efle eut det~cting-program according to the present invention. In the preferred embodiment ofthe pTesent invention, the value of k in Eq. (1) is set to one. To sirnplify the notation, Pn I has been replaced by p in the following discussion. After initializing various variables as 10 shown at 10, the program acquires the next frame of the motion picture and computes the values for p and the average value of ~J as shown at 12. The program then tests the value of p to deterrnine if an abrupt cut has occurred as shown at 14. An abrupt cut is deemed to have occurred if the value of p is less than a predetermined threshold. It is possible for an abrupt cut to take more than one frame. Hence, a flag l S is set as shown at 16 to indicate that p has crossed the threshold. The flag will be reset when p once again retums to a value above the threshold. This flag is used by blocks 18 and 20 to determine if the current value of p being less than the threshold is a new cut or just a continuation of the previously detected abrupt cut. If the cut is completed, the flag is reset as shown at 22.
If no abrupt cut is in progress, the program goes on to examine the data to determine if a gradual cut is in progress. As noted above, the program uses the average value of <~ to detect gradual cuts. The curve consisting of the average value of ~ as a function of the frame number will be referred to as the ~s curve. The 25 program also keeps track of the slope of the cJ curve. Changes in the slope of the a curve are used to detect possible events in traversing a gradual cut. As noted above, a gradual cut may be evidenced either by a peak in the ~ curve or by a ramp connecting two ~ values representative of ~s in the frarnes on each side of the gradual cut. In the later case, there will be no change in the sign of the slope of the ~ curve during the 30 gradual cut.
Referring again to Figure 1, if no abrupt cut is in progress, the program tests for a gradual cut being in progress as shown at 24. If a gradual cut is not in progress, the program checks for the beginning of a gradual cut as shown at 26. A gradual cut 2 ~ 2 is deemed to have begun if ~e slope of ~e ~ curve has changed and ~e change in 6is greater than a predetennined threshold value. If these conditions are met, a set of flags is initialized as shown at 28. The flag values indicate that a gradual cut has begun and no peak has yet been detected in ~e 6 curve.
If ~a~al cut-was already-in progres~-a~ ate~by-the flags set at 28, the program determines if the current f~ame is the end of the gradual cut as shown at 30. -If the sign of the slope of the c~ curve has changed, the program checks to see if the program has already passed through an inflection point as indicated by one of the 10 above-described flags having a value different from its initialization value. If the initial value is still present, the program assumes that the change in sign is due to the presence of an inflection point in the ~ curve. The flag is set accordingly as shown at 32. If the ~ curve has already passed through an inflection point, the program assumes that the gradual cut has been completed.
If the sign of the slope of the ~ curve has not changed but the change in the c~curve from the last frame is less than some predetennined threshold, the programassumes that the gradual cut has been completed, provided the excursions in the ~
curve from one scene to the next were sufficiently large. This case is encountered 20 when there is no peak in the a curve between the scenes.
The a curve, even with smoothing, has a significant amount of noise. To distinguish a "ramp" between two values in the ~s curve from noise, the program keeps track of certain reference values on the <s curve. The value of the a curve at the 25 potential beginning of a ramp is stored. At the end of the ramp, the difference between the ending value and the reference values is used to eliminate "noise events".
While the preferred embodiment of the present invention uses the standard deviation of the pixel values in each frame for detecting gradual cuts, it will be 30 apparent to those skilled in the art that other statistical measures or functions thereof may be used. In this regard, it should be noted that ~ is a function of the variance of the pixel values which in turn is the average value of the second moment of the pixel values. Hence, any function based on one of the moments of the pixel distribution may also be utilized.
~ ~ ~ r :~
2 ~ 2 -7~
While the above-described embodiments of the present invention compute the correlation and standard deviation using all of the data in each frame, it w~ll be apparent to those skilled in the art that a subset of the pixel values may be used to determine these parameters. Any representative subset may be used. Embodiments ~ -based on such subsets generate a lower computational load; however, such embodiments are expected to yield ~ values having more "noise".
The present invention is preferably practiced on a general purpose computer.
However, it will be apparent to those skilled in the art that special purpose computing hardware may also h utilized to improve the speed with which the method operates.
For example, hardware that computes the correlation of t~vo frames and the variance of the pixels in a given frame can substantially improve ~e speed with which the ~ `
method of the present invention operates. -Various modifications to the present invention will become apparent to those skilled in the art from the foregoing description and accompanying drawings.
Accordingly, the present invention is to be limited solely by the scope of the following claims. ~ ; -~ ~ S
Claims (7)
1. A method for operating a computer to detect scene cuts in a motion picture, said motion picture comprising an ordered sequence of frames, each said frame comprising a plurality of intensity values, said method comprising the steps of determining a correlation value of each frame with the a corresponding previous frame, said previous frame being separated in said sequence by a predetermined number of frames;
comparing said correlation value with a threshold value; and reporting an abrupt scene change if said correlation value is less than said threshold value.
comparing said correlation value with a threshold value; and reporting an abrupt scene change if said correlation value is less than said threshold value.
2. The method of Claim I wherein said predetermined number of frames is equal to one.
3. The method of Claim I further comprising the steps of determining a statistical measure for each said frame; and detecting changes in the value of said statistical measure as a function of the frame number and;
reporting a gradual scene change if said detected changes match a predetermined pattern.
reporting a gradual scene change if said detected changes match a predetermined pattern.
4. The method of Claim 3 wherein said statistical measure is a function of the variance of said intensity values.
5. The method of Claim 3 wherein said statistical measure is a moment of said intensity values.
6. The method of Claim 3 wherein said predetermined pattern is a peak in said statistical measure as a function of the frame number.
7. The method of Claim 3 wherein said predetermined pattern is a ramp connecting two regions in said function, said regions of a variation of said statistical measure less than the difference in said statistical measure between the beginning and end of said ramp.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US9911393A | 1993-07-29 | 1993-07-29 | |
| US099,113 | 1993-07-29 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CA2114052A1 true CA2114052A1 (en) | 1995-01-30 |
Family
ID=22272820
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA 2114052 Abandoned CA2114052A1 (en) | 1993-07-29 | 1994-01-24 | Method of detecting scene cuts |
Country Status (3)
| Country | Link |
|---|---|
| EP (1) | EP0637027A3 (en) |
| JP (1) | JPH0837621A (en) |
| CA (1) | CA2114052A1 (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3823333B2 (en) * | 1995-02-21 | 2006-09-20 | 株式会社日立製作所 | Moving image change point detection method, moving image change point detection apparatus, moving image change point detection system |
| GB9607645D0 (en) * | 1996-04-12 | 1996-06-12 | Snell & Wilcox Ltd | Processing of video signals prior to compression |
| DE19637139C1 (en) * | 1996-09-12 | 1997-12-04 | Philips Patentverwaltung | Video signal processing device with image error detection |
| DE19637138C1 (en) * | 1996-09-12 | 1997-12-04 | Philips Patentverwaltung | Video signal processing device with phantom image detection |
| US6084641A (en) * | 1997-08-06 | 2000-07-04 | General Instrument Corporation | Fade detector for digital video |
| FR2779549B1 (en) | 1998-06-08 | 2000-09-01 | Thomson Csf | METHOD FOR SEPARATING THE DYNAMIC AND STATIC COMPONENTS OF A SUITE OF IMAGES |
| US6493042B1 (en) * | 1999-03-18 | 2002-12-10 | Xerox Corporation | Feature based hierarchical video segmentation |
| JP2001292372A (en) * | 2000-04-04 | 2001-10-19 | Pioneer Electronic Corp | Image change detector, image change detecting method, image encoder and information recording medium in which program for detecting image change is computer- readably recorded |
| GB2362533A (en) | 2000-05-15 | 2001-11-21 | Nokia Mobile Phones Ltd | Encoding a video signal with an indicator of the type of error concealment used |
| GB2362531A (en) | 2000-05-15 | 2001-11-21 | Nokia Mobile Phones Ltd | Indicating the temporal order of reference frames in a video sequence |
| GB2362532B (en) | 2000-05-15 | 2004-05-05 | Nokia Mobile Phones Ltd | Video coding |
| US20070201746A1 (en) * | 2002-05-20 | 2007-08-30 | Konan Technology | Scene change detector algorithm in image sequence |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2977829B2 (en) * | 1989-01-11 | 1999-11-15 | 株式会社東芝 | Moving picture reproducing apparatus and moving picture reproducing method |
| US5099322A (en) * | 1990-02-27 | 1992-03-24 | Texas Instruments Incorporated | Scene change detection system and method |
| JP2863818B2 (en) * | 1990-08-31 | 1999-03-03 | 工業技術院長 | Moving image change point detection method |
| JP3067275B2 (en) * | 1991-05-23 | 2000-07-17 | 日本放送協会 | Scene change detection device |
| FR2683415B1 (en) * | 1991-10-30 | 1996-08-09 | Telediffusion Fse | SYSTEM FOR VIDEO ANALYSIS OF THE ASSEMBLY OF A BROADCASTED OR RECORDED TELEVISION PROGRAM AND ITS USE FOR POST PRODUCTION TECHNIQUES, ESPECIALLY MULTILINGUAL. |
| JPH06339109A (en) * | 1993-05-28 | 1994-12-06 | Matsushita Electric Ind Co Ltd | Scene change detection method, its device and moving picture edit device |
-
1994
- 1994-01-24 CA CA 2114052 patent/CA2114052A1/en not_active Abandoned
- 1994-07-25 EP EP94305571A patent/EP0637027A3/en not_active Withdrawn
- 1994-07-29 JP JP19729194A patent/JPH0837621A/en active Pending
Also Published As
| Publication number | Publication date |
|---|---|
| JPH0837621A (en) | 1996-02-06 |
| EP0637027A2 (en) | 1995-02-01 |
| EP0637027A3 (en) | 1995-11-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US5245436A (en) | Method and apparatus for detecting fades in digital video sequences | |
| CA2114052A1 (en) | Method of detecting scene cuts | |
| EP1048169B1 (en) | Detection of transitions in video sequences | |
| US6061471A (en) | Method and system for detecting uniform images in video signal | |
| Fernando et al. | Fade and dissolve detection in uncompressed and compressed video sequences | |
| DE69714224T2 (en) | METHOD AND DEVICE FOR REDUCING BLOCK EFFECTS IN IMAGES | |
| EP0690413A2 (en) | A system for locating automatically video segment boundaries and for extraction of key-frames | |
| US6101222A (en) | Scene change detection | |
| DE60023779T2 (en) | Apparatus and method for fade detection and fade information coding | |
| US6810145B2 (en) | Process for detecting a change of shot in a succession of video images | |
| US20030112874A1 (en) | Apparatus and method for detection of scene changes in motion video | |
| EP1914994A1 (en) | Detection of gradual transitions in video sequences | |
| EP0830018A3 (en) | Method and system for motion detection in a video image | |
| US5544239A (en) | Method and apparatus for improving motion analysis of fades | |
| CN110889328A (en) | Method, device, electronic equipment and storage medium for detecting road traffic condition | |
| JP3173834B2 (en) | Method and apparatus for segmenting a signal representing a digital image | |
| US7982774B2 (en) | Image processing apparatus and image processing method | |
| US6707943B2 (en) | Method of monitoring the quality of distributed digital images by detecting false contours | |
| US6459459B1 (en) | Method for detecting transitions in sampled digital video sequences | |
| KR20010060316A (en) | Detection of gaussian noise in video signals | |
| US5283645A (en) | Method and apparatus for detecting dissolve regions in digital video sequences | |
| EP1973351B1 (en) | Monitor | |
| US8270483B1 (en) | Method and apparatus for applying accurate parameter estimation and efficient fade detection in weighted prediction for video compression | |
| JP2859345B2 (en) | Scene change detection method | |
| US20060268181A1 (en) | Shot-cut detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| FZDE | Dead |